Stedi Interoperability · Engineering Field Note

The Clearinghouse Is Not the Payer Platform

Cloud Health Office is the claims operating system a health plan actually runs. Stedi is one of the networks it can speak.

Evidence scope: This field note describes merged, inspectable pull requests on a vendor-neutral healthcare transaction gateway. It is not a production-volume result, not a live ERA enrollment certificate, and not payment posting. Live Stedi 835 retrieve still needs payer enrollment and a real transaction id. Scale evidence lives in the Million Claim Challenge.
Inspect the PRs Gateway architecture Article source MCC evidence
Dark industrial hall with a glowing cyan monolith at the center and two fiber spines carrying sealed capsules toward and away from it
The network carries transactions. The platform decides what they mean. Cloud Health Office sits at the core; Stedi is a spine into the rest of the industry.

The question investors actually ask

Healthcare investors do not diligence a clearinghouse the way they diligence a payer platform. A pipe that can post an 837 is necessary and nowhere near sufficient. The strict criteria usually sound like this:

That last question is the trap. Wrap Stedi too tightly and you have rented a network, not built a platform. Rebuild QNXT and you have a ten-year migration. Cloud Health Office is the third shape: a source-available claims platform with a replaceable network adapter.

Eight-panel diagram mapping investor criteria: eligibility, claims, follow-through, intelligence, tenant and PHI, compliance, scale proof, and vendor risk
The diligence checklist, mapped onto shipped CHO surfaces. Stedi is the first implemented network. The gateway is the contract that survives a vendor swap.

Ninety seconds on what CHO is

Cloud Health Office is a Kubernetes-native, multi-tenant claims administration platform. It does the work a health plan actually has to do after the EDI file arrives: interpret eligibility, ingest the claim, adjudicate against a versioned benefit plan, price against fee schedules, apply NCCI and COB, persist a lifecycle, and expose FHIR APIs for CMS-0057-F.

It is source-available under BSL 1.1. It deploys in the plan’s cloud. It is designed to sit beside QNXT, Facets, or HealthEdge as a compliance and modernization layer, or to run as a full claims stack for a new entrant.

The Million Claim Challenge already showed the adjudication core can carry a million mixed claims through Azure Service Bus on a local Kubernetes cluster, with inspectable evidence. This field note is the other half of the same argument: the platform can also talk to the industry without becoming the industry’s clearinghouse.

Stedi moves HIPAA transactions. Cloud Health Office owns eligibility interpretation, claim identity, acknowledgment, status, attachments, remittance, and the workflow view a provider or dental application can actually use.

The boundary that makes the platform valuable

Every Stedi PR in this sequence obeyed one architectural rule: CHO owns the healthcare business. The gateway owns transport and translation. Canonical request and response models never contain a vendor name. A guard test fails the build if they do.

That sounds academic until you price vendor lock-in. A health plan that embeds Stedi DTOs in its claim aggregate cannot change networks without rewriting adjudication. A health plan that talks to IClaimSubmissionGateway can add Availity later the way CHO added Stedi: an adapter, not a product rewrite.

Layered architecture diagram showing CHO business services above a vendor-neutral gateway, with Stedi implemented and Availity or direct payer as later adapters
Layer 1 is the healthcare business CHO already owns. Layer 2 is the vendor-neutral gateway. Stedi is the first implemented pipe under that gateway, not the product itself.

One claim life, six transactions

A real claim is not one API call. It is a life:

  1. 270/271 — is the member covered?
  2. 837 — submit professional, institutional, or dental.
  3. 277CA — accepted or rejected into processing. Not paid.
  4. 276/277 — where is it now? Status is not remittance.
  5. 275 — outbound documents, and inbound receive when CHO is the payer.
  6. 835 — what was allowed, paid, and left to the patient. Stored, not posted.

Most “EDI platforms” collapse those into a single status field. CHO keeps them distinct on purpose. That is how you avoid the lie that kills trust with providers: calling a clearinghouse acceptance a payment.

Six glowing instrument stations along a cyan conveyor, representing sequential healthcare transactions in a dark operations hall
Each HIPAA transaction is a station, not an overwrite. Eligibility, submission, acknowledgment, status, attachments, and remittance keep their own durable records.
Six labeled stations for 270/271, 837, 277CA, 276/277, 275, and 835 feeding a claim intelligence API
The spine that just landed in main. Claim intelligence sits underneath the transactions, composing them rather than replacing them.
TransactionBusiness questionEvidence
270 / 271Is this member covered?Stedi eligibility JSON adapter
837P / I / DCan we send this claim?PR #1111
277CADid processing accept it?#1112, #1114
275Does the payer need more?#1115 outbound, #1116 inbound
276 / 277Where is it now?PR #1117
835What was paid?PR #1118
IntelligenceWhat should we do next?PR #1119
6HIPAA transaction types on one gateway
9sequential PRs from 837 through intelligence
0vendor names allowed in canonical models
1tenant-safe read API for CDO and portal

Claim intelligence is the product surface

A dental practice, a provider portal, and a payer operations queue should not have to understand CLP02. They need a sentence: processing, pending information, paid, denied. They need a timeline. They need to know whether documentation arrived. They need patient responsibility when an ERA exists, and “unknown” when it does not.

GET /api/claims/{claimId}/intelligence is that sentence. It rebuilds from the transaction stores. It is not a second database of truth. Duplicate 277CA or 835 deliveries do not duplicate the timeline, because event ids are the source record ids — status lives on the payload, not in the identity.

Control room overlooking a river of claim packets, with operators facing a constellation wall and vaulted drawers of transaction layers below
The product a provider or dental application actually wants: one observatory over the claim, not six EDI manuals.
Diagram of 837, 277CA, 276/277, 275, and 835 stores feeding a composer consumed by CloudDentalOffice, portal, operations, and future AI
CloudDentalOffice, a future provider portal, operations, and later AI consume the same composer. None of them should re-implement HIPAA.

Two rules from that PR are the ones to say out loud in a diligence room:

Those are not slogans. They are tests. They are why a healthcare investor can believe the rest of the lifecycle.

What we refused to fake

The Million Claim Challenge series is useful here as a method, not just a number. We publish limitations in the same breath as results. For Stedi:

That list is how you tell a serious healthcare investor you are not selling a demo that collapses in underwriting.

Why this is the valuable platform

The value is not “we integrated Stedi.” Plenty of teams can post an 837. The value is the combination:

  1. A complete claim life — eligibility through remittance — as distinct facts.
  2. A healthcare business layer that already adjudicates, prices, and scores at million-claim evidence scale.
  3. A replaceable network so the plan is not married to one clearinghouse.
  4. A workflow API that CloudDentalOffice and a provider portal can use without becoming EDI companies.
  5. CMS-0057-F FHIR on the same platform, so compliance is an on-ramp rather than a side project.
  6. Source you can read. The PRs, the evidence docs, and the Million Claim Challenge packets are public.

For a capital-markets conversation, that is the underwriting story. The clearinghouse is a cost of doing business. The platform is the asset: tenant-safe claims operations that a health plan, a TPA, or a dental network can run, inspect, and extend.

Stedi made the wire real this month. Cloud Health Office is what the wire is for.

Stedi series Million Claim Challenge series Claim intelligence architecture Schedule a discovery call